{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://ai-shifu.local/schemas/course-direction-advisor.v1.json",
  "title": "Course Direction Advisor Output",
  "type": "object",
  "additionalProperties": false,
  "required": [
    "meta",
    "material_core",
    "target_segments",
    "market_judgment",
    "directions",
    "decision",
    "evidence_appendix"
  ],
  "properties": {
    "meta": {
      "type": "object",
      "additionalProperties": false,
      "required": [
        "task_id",
        "created_at",
        "analysis_mode",
        "report_language",
        "priority_markets"
      ],
      "properties": {
        "task_id": { "type": "string", "minLength": 1 },
        "created_at": { "type": "string", "format": "date-time" },
        "analysis_mode": {
          "type": "string",
          "enum": ["A-material-only", "B-market-scan"]
        },
        "report_language": { "type": "string", "minLength": 2 },
        "priority_markets": {
          "type": "array",
          "minItems": 1,
          "items": { "type": "string", "minLength": 2 }
        }
      }
    },
    "material_core": {
      "type": "object",
      "additionalProperties": false,
      "required": [
        "theme_cues",
        "problem_cues",
        "outcome_cues",
        "proof_cues",
        "boundary_cues"
      ],
      "properties": {
        "theme_cues": { "type": "array", "items": { "$ref": "#/$defs/evidence_item" } },
        "problem_cues": { "type": "array", "items": { "$ref": "#/$defs/evidence_item" } },
        "outcome_cues": { "type": "array", "items": { "$ref": "#/$defs/evidence_item" } },
        "proof_cues": { "type": "array", "items": { "$ref": "#/$defs/evidence_item" } },
        "boundary_cues": { "type": "array", "items": { "$ref": "#/$defs/evidence_item" } }
      }
    },
    "target_segments": {
      "type": "array",
      "minItems": 1,
      "maxItems": 3,
      "items": {
        "type": "object",
        "additionalProperties": false,
        "required": [
          "segment_name",
          "who_they_are",
          "current_state",
          "job_to_be_done",
          "purchase_motivation",
          "willingness_to_pay",
          "current_alternatives",
          "buying_driver",
          "source_ref"
        ],
        "properties": {
          "segment_name": { "type": "string", "minLength": 1 },
          "who_they_are": { "type": "string", "minLength": 1 },
          "current_state": { "type": "string", "minLength": 1 },
          "job_to_be_done": { "type": "string", "minLength": 1 },
          "purchase_motivation": { "type": "string", "minLength": 1 },
          "willingness_to_pay": {
            "type": "string",
            "enum": ["high", "medium", "low", "unknown"]
          },
          "budget_band": { "type": "string" },
          "current_alternatives": {
            "type": "array",
            "minItems": 1,
            "items": { "type": "string", "minLength": 1 }
          },
          "buying_driver": {
            "type": "array",
            "minItems": 1,
            "items": {
              "type": "string",
              "enum": ["trust", "relevance", "executability", "value"]
            }
          },
          "source_ref": { "$ref": "#/$defs/ref_array" }
        }
      }
    },
    "market_judgment": {
      "type": "object",
      "additionalProperties": false,
      "required": [
        "lifecycle_stage",
        "competition_type",
        "market_summary",
        "competition_matrix",
        "opportunity_gap"
      ],
      "properties": {
        "lifecycle_stage": {
          "type": "string",
          "enum": [
            "information-explosion",
            "segmented-understanding",
            "methodology-phase",
            "toolification-phase",
            "red-ocean"
          ]
        },
        "competition_type": {
          "type": "array",
          "minItems": 1,
          "items": {
            "type": "string",
            "enum": ["content", "tool", "service"]
          }
        },
        "market_summary": { "type": "string", "minLength": 1 },
        "competition_matrix": {
          "type": "array",
          "minItems": 3,
          "items": {
            "type": "object",
            "additionalProperties": false,
            "required": [
              "name",
              "solution_type",
              "problem_solved",
              "target_user",
              "why_buyers_choose_it",
              "strengths",
              "what_it_does_not_solve",
              "opportunity_for_us",
              "source_link"
            ],
            "properties": {
              "name": { "type": "string", "minLength": 1 },
              "solution_type": {
                "type": "string",
                "enum": ["content", "tool", "service", "platform-feature", "free-content", "training-camp"]
              },
              "problem_solved": { "type": "string", "minLength": 1 },
              "target_user": { "type": "string", "minLength": 1 },
              "why_buyers_choose_it": { "type": "string", "minLength": 1 },
              "strengths": {
                "type": "array",
                "minItems": 1,
                "items": { "type": "string", "minLength": 1 }
              },
              "what_it_does_not_solve": { "type": "string", "minLength": 1 },
              "opportunity_for_us": { "type": "string", "minLength": 1 },
              "source_link": { "type": "string", "format": "uri" },
              "evidence_tier": {
                "type": "string",
                "enum": ["A", "B", "C"]
              }
            }
          }
        },
        "opportunity_gap": { "type": "string", "minLength": 1 }
      }
    },
    "directions": {
      "type": "object",
      "additionalProperties": false,
      "required": ["primary", "backups", "not_recommended"],
      "properties": {
        "primary": { "$ref": "#/$defs/direction_item" },
        "backups": {
          "type": "array",
          "minItems": 2,
          "maxItems": 3,
          "items": { "$ref": "#/$defs/direction_item" }
        },
        "not_recommended": {
          "type": "array",
          "minItems": 1,
          "items": {
            "type": "object",
            "additionalProperties": false,
            "required": ["title", "reason", "risk"],
            "properties": {
              "title": { "type": "string", "minLength": 1 },
              "reason": { "type": "string", "minLength": 1 },
              "risk": { "type": "string", "minLength": 1 }
            }
          }
        }
      }
    },
    "decision": {
      "type": "object",
      "additionalProperties": false,
      "required": [
        "decision_status",
        "decision_reason",
        "course_readiness_summary",
        "top_risks",
        "scorecard",
        "price_hint",
        "mvp_scope",
        "validation_plan_7d"
      ],
      "properties": {
        "decision_status": {
          "type": "string",
          "enum": ["GO", "HOLD", "REWORK", "NO-GO"]
        },
        "decision_reason": { "type": "string", "minLength": 1 },
        "course_readiness_summary": { "type": "string", "minLength": 1 },
        "top_risks": {
          "type": "array",
          "minItems": 3,
          "items": { "$ref": "#/$defs/risk_item" }
        },
        "scorecard": {
          "type": "object",
          "additionalProperties": false,
          "required": [
            "source_strength",
            "user_problem_fit",
            "market_opportunity",
            "differentiation_strength",
            "commercialization_feasibility",
            "total"
          ],
          "properties": {
            "source_strength": { "type": "number", "minimum": 0, "maximum": 25 },
            "user_problem_fit": { "type": "number", "minimum": 0, "maximum": 20 },
            "market_opportunity": { "type": "number", "minimum": 0, "maximum": 20 },
            "differentiation_strength": { "type": "number", "minimum": 0, "maximum": 20 },
            "commercialization_feasibility": { "type": "number", "minimum": 0, "maximum": 15 },
            "total": { "type": "number", "minimum": 0, "maximum": 100 }
          }
        },
        "price_hint": { "type": "string", "minLength": 1 },
        "mvp_scope": {
          "type": "array",
          "minItems": 1,
          "items": { "type": "string", "minLength": 1 }
        },
        "validation_plan_7d": {
          "type": "array",
          "minItems": 1,
          "items": { "type": "string", "minLength": 1 }
        }
      }
    },
    "evidence_appendix": {
      "type": "object",
      "additionalProperties": false,
      "required": ["claim_evidence_map", "market_links", "missing_info"],
      "properties": {
        "claim_evidence_map": {
          "type": "array",
          "items": {
            "type": "object",
            "additionalProperties": false,
            "required": ["claim", "refs"],
            "properties": {
              "claim": { "type": "string", "minLength": 1 },
              "refs": { "$ref": "#/$defs/ref_array" }
            }
          }
        },
        "market_links": {
          "type": "array",
          "items": { "type": "string", "format": "uri" }
        },
        "missing_info": {
          "type": "array",
          "items": { "type": "string", "minLength": 1 }
        }
      }
    }
  },
  "$defs": {
    "ref_array": {
      "type": "array",
      "minItems": 1,
      "items": {
        "type": "string",
        "pattern": "^(SRC-[A-Z0-9-]+|MKT-[A-Z0-9-]+|https?://.+)$"
      }
    },
    "evidence_item": {
      "type": "object",
      "additionalProperties": false,
      "required": ["statement", "refs", "confidence"],
      "properties": {
        "statement": { "type": "string", "minLength": 1 },
        "refs": { "$ref": "#/$defs/ref_array" },
        "confidence": {
          "type": "string",
          "enum": ["high", "medium", "low"]
        }
      }
    },
    "direction_item": {
      "type": "object",
      "additionalProperties": false,
      "required": [
        "title",
        "topic_type",
        "target_user",
        "problem_solved",
        "teaching_goal",
        "one_line_pitch",
        "why_this_direction",
        "fit_reason",
        "go_to_market_hint",
        "growth_shape",
        "expected_curve",
        "upside_case",
        "downside_case",
        "course_readiness",
        "course_readiness_reason",
        "lesson_spine",
        "teachable_kernel",
        "source_ref"
      ],
      "properties": {
        "title": { "type": "string", "minLength": 1 },
        "topic_type": {
          "type": "string",
          "enum": ["cognitive", "method", "case", "execution", "outcome-led", "hybrid"]
        },
        "target_user": { "type": "string", "minLength": 1 },
        "problem_solved": { "type": "string", "minLength": 1 },
        "teaching_goal": { "type": "string", "minLength": 1 },
        "one_line_pitch": { "type": "string", "minLength": 1 },
        "why_this_direction": { "type": "string", "minLength": 1 },
        "fit_reason": { "type": "string", "minLength": 1 },
        "go_to_market_hint": { "type": "string", "minLength": 1 },
        "growth_shape": {
          "type": "string",
          "enum": ["trend-led", "durable-demand", "hybrid"]
        },
        "expected_curve": {
          "type": "string",
          "enum": ["spike", "steady-compound", "hybrid", "uncertain"]
        },
        "upside_case": { "type": "string", "minLength": 1 },
        "downside_case": { "type": "string", "minLength": 1 },
        "course_readiness": {
          "type": "string",
          "enum": ["course-ready", "partially-course-ready", "topic-ready-only", "insufficient"]
        },
        "course_readiness_reason": { "type": "string", "minLength": 1 },
        "lesson_spine": {
          "oneOf": [
            {
              "type": "string",
              "enum": ["insufficient"]
            },
            {
              "type": "array",
              "minItems": 3,
              "items": { "type": "string", "minLength": 1 }
            }
          ]
        },
        "teachable_kernel": {
          "type": "array",
          "minItems": 1,
          "items": { "type": "string", "minLength": 1 }
        },
        "source_ref": { "$ref": "#/$defs/ref_array" }
      }
    },
    "risk_item": {
      "type": "object",
      "additionalProperties": false,
      "required": ["name", "severity", "trigger", "evidence", "decision_impact"],
      "properties": {
        "name": { "type": "string", "minLength": 1 },
        "severity": {
          "type": "string",
          "enum": ["low", "medium", "high"]
        },
        "trigger": { "type": "string", "minLength": 1 },
        "evidence": { "type": "string", "minLength": 1 },
        "decision_impact": { "type": "string", "minLength": 1 }
      }
    }
  }
}

# Decision Model

## Core Idea

The job is not to simply name a course. The job is to complete a commercially useful translation under two hard constraints:

- content reality: stay inside the material
- market reality: stay inside real demand and real competition

Failure usually comes from satisfying only one side:

- content-only thinking produces topics that sound thoughtful but do not sell
- market-only thinking produces topics that sound sellable but the author cannot truly teach

The correct target is:

- `Minimum Sellable Topic`

## What Counts as a Minimum Sellable Topic

A direction qualifies only if all four are true:

- the material has a stable content core
- the market shows a recognizable demand
- a specific audience could plausibly pay for the outcome or efficiency gain
- the author or material has enough support to carry the delivery

If one of these is missing, do not finalize the topic.

## How Far Commercial Reframing Is Allowed

### Allowed

- expression reframing: clarify what was previously loose or scattered
- problem mapping: turn “what the author talks about” into “what the user is trying to solve”
- audience narrowing: reduce a broad audience into a sharper segment
- result narrowing: pull large or vague outcomes back to what the material truly supports
- product downgrading: shift from full course to lighter product when necessary

### Not Allowed

- content invention
- evidence inflation
- result inflation
- authority inflation

## The Five Required Judgments

### 1. Content Compression

Answer:

- what the material consistently covers
- what problem it is genuinely good at addressing
- what themes, moves, cases, and judgments recur
- what larger claims it does not support

### 2. User Mapping

Find the overlap between:

- people who need this
- people likely to pay for this

For each serious audience candidate, answer:

- current state
- desired outcome
- what they do not want to do
- what they might pay for
- why they would trust the author
- why they might still not buy

### 3. Market Positioning

Answer:

- current market stage
- dominant solution types
- competitive density and pattern
- what remains insufficiently solved

### 4. Sellable Packaging

Do not default to a full course.

Possible packaging modes:

- cognitive
- method
- case
- execution
- outcome-led
- hybrid

The evidence ceiling determines the packaging ceiling.

### 5. Validation Strength

Judge whether the direction is:

- merely logical
- or already meaningfully validated by public signals

Ask:

- is there recurring discussion, not just a spike
- are users repeatedly asking for templates, examples, execution help, or clarification
- is there stable supply of comparable solutions
- is this a durable demand or a short-term framing spike
- does the author's real strength match a real market gap

## When a Course Topic Should Be Rejected

Any of the following is enough to reject or downgrade a course direction:

- the material is too scattered to form a stable problem core
- demand exists but the material does not contain a matching solution
- market attention exists but competition is saturated and the material has no real edge
- the material supports only understanding, but the audience pays for outcomes
- the author cannot support the promise level
- the heat comes from a short event and the material does not convert into a durable problem
- the segment is already crowded and the material lacks a people / scenario / evidence edge

Rejecting the course direction is a valid result, not a failure of the analysis.

## Downgrade Logic

If a full course does not hold, test whether the material can support:

- a case breakdown
- an experience share
- a conceptual clarification
- a method fragment
- an observation report

Even a downgraded direction must still answer:

- who it is for
- why they would care
- why they might buy it

## Buying Logic

Most buyers respond to some combination of:

- `trust`
- `relevance`
- `executability`
- `value`

The report should specify which of these matters most for each serious audience segment, and whether the material can support that purchase logic.

## Recommended Order of Operations

1. compress the material into a stable capability core
2. list candidate users and anti-users
3. derive search terms and alternative phrasing from the material
4. scan public discussion, paid products, complaints, and substitutes
5. judge stage, density, replacement risk, and trend pattern
6. compare the material's edge with the market gap
7. output a recommendation, backup directions, downgrade, or rejection

# Decision Report Template

Use this template when the user needs a decision-oriented course-topic report rather than a loose market summary.

The report should help the user decide:

- which direction to push
- why that direction wins
- what the backup routes are
- why other routes should not be prioritized
- how the direction should be sold, priced, and scoped

## When to Use This Template

Use this template only when:

- the material has already been compressed into a stable capability core
- a basic market scan is complete
- you can propose at least one main direction and two meaningful backups

If the material or market evidence is too weak, downgrade to a conservative feasibility memo rather than forcing a full decision report.

## Analysis Mode

Always state the mode:

- `A-material-only`
- `B-market-scan`

If the user asks in a specific language:

- write the report in that language unless they ask otherwise
- include and prioritize markets aligned with that language

## Lifecycle Labels

Use the canonical lifecycle set:

- `information-explosion`
- `segmented-understanding`
- `methodology-phase`
- `toolification-phase`
- `red-ocean`

## Core Principles

Always preserve these rules:

- material boundary first, market amplification second
- promise ceiling must not exceed the evidence ceiling
- market size, sales potential, and breakout potential must remain evidence-constrained judgments
- commercial framing is allowed; fabricated advantage is not

## Recommended Structure

### 1. Conclusion First

The first screen should give:

- one recommended direction
- two backup directions
- two to three directions that should not be pursued

For each recommended direction, include at least:

- what kind of course it is
- who it is for
- what problem it solves
- whether it is more of a trend-led or durable-demand play

### 2. Direction Overview

Compare all serious directions in one place.

At minimum include:

- direction name
- course type
- target audience
- problem solved
- rough market room
- growth shape
- expected growth curve
- lifecycle stage

The main and backup directions must be genuinely different.

They should differ on at least one of:

- target audience
- problem framing
- teaching objective
- product form

If they are merely title variants, treat them as the same direction.

### 3. Why the Decision Was Reached

Explain:

- what user-problem cluster was identified in the market
- how the source material maps back to that cluster
- why the recommended direction beats the alternatives
- whether the material contains enough coherent knowledge substance to support a real course

This section should make the reasoning obvious, not just the conclusion.

### 4. Competitor and Substitute Analysis

Analyze at least three competitors or substitutes from different categories:

- method/content products
- tools or platform features
- services, training camps, or free-content substitutes

For each one, explain:

- what problem it solves
- who it serves
- why users buy it
- its main strength
- what it does not solve
- what opportunity remains for this topic
- what distinctive teachable kernel the source material has, if any, compared with that competitor
- source link

Do not dump links without interpretation.

### 5. Market Stage and Buying Logic

Explain plainly:

- the current stage of the market
- what users now need
- what they are actually willing to pay for
- whether the dominant substitute is content, tool, or service

### 5.5 Growth Curve

For each serious direction, explain:

- whether it is more likely to spike, compound steadily, or follow a hybrid curve
- what would make the direction grow well
- what would make it stall or fade

This helps prevent the report from treating all “good topics” as commercially identical.

### 6. Directions Not Recommended

List at least two directions that should not be pursued.

Typical reasons:

- outside the material boundary
- clearly crowded
- low willingness to pay for a course
- better solved by a tool or service
- driven by a short-term framing spike rather than durable demand

### 7. Final Recommendation

For the recommended direction, specify:

- title
- teaching objective
- course type
- recommended price band
- main selling points
- why the material is course-ready rather than only topic-ready
- the minimum lesson spine, normally at least three meaningful lessons
- minimum launch scope

For backups, specify:

- title
- audience
- teaching objective
- why the direction ranks below the winner

### 8. Author or Team Advantage

If the material supports a real edge, state it explicitly:

- real background
- representative proof
- scarce perspective
- delivery strength
- specific methods, cases, experience, or insight density that competitors are less likely to have

The goal is to explain why this author has a right to teach this topic.

### 9. Decision Gate

Add:

- decision status
- decision reason
- top risks
- a plain judgment on lesson-level sufficiency and internal coherence

The report should feel like a decision memo, not a brainstorm note.

### 10. MVP and Seven-Day Validation Plan

Include:

- the smallest launchable version
- the initial pricing hypothesis
- a seven-day validation plan

Typical validation steps:

- interview target users
- scan comparable offers
- test three value propositions
- produce one sample lesson or sample asset
- review feedback and refine positioning

### 10.5 Optional Weighted Scorecard

If a sharper decision gate is useful, add a weighted scorecard:

- source strength: 25
- user-problem fit: 20
- market opportunity: 20
- differentiation strength: 20
- commercialization feasibility: 15

Use the scorecard as a support tool, not as a substitute for reasoning.

### 11. Evidence Appendix

Add:

- `claim_evidence_map`
- external links
- missing-evidence checklist

Also include a short `course-readiness check`:

- whether the material can support at least three lessons
- what the lesson spine would be
- what is still missing if the topic is only partially course-ready

## Readability Rules

If the report is meant for human decision-making:

- make the conclusion readable in under one minute
- make the reasoning readable in the next section
- avoid field-name-heavy presentation unless the user explicitly wants a structured version
- avoid vague references such as “this direction” when the full topic name would be clearer
- make the report read like a decision document, not an interface payload

# Gating Rules

## Must-Pass Rules

All of the following must hold:

1. Every recommended topic is traceable to source evidence.
2. Every target audience is grounded in the material, not invented from intuition.
3. Every market claim is supported by current public evidence.
4. Every course promise stays within the source-evidence ceiling.
5. Every claimed differentiator comes from the material, not generic marketing language.
6. If the material supports only understanding, do not package it as a high-promise outcome product.
7. If user groups differ too much, do not force them into one course.
8. If the space is crowded and the material has no real edge, say so explicitly.
9. If evidence is missing, list the missing inputs clearly instead of filling them in silently.
10. The final recommendation must explain why this topic wins over larger or hotter alternatives.
11. Distinguish people who need the topic from people who will pay for the course.
12. If the buyer is more likely to choose a tool or service, say so explicitly.
13. If no minimum sellable topic exists, a `not-viable` result is valid.
14. If a full course does not hold, test downgrade paths first.
15. Market size can only be a public-signal estimate, never fake precision.
16. Trend judgments must compare multiple related terms, not one hot keyword.
17. One viral post or one spike is not enough to prove course demand.
18. SEO comparison is limited to public content surfaces, not internal analytics.
19. If the heat sits in the framing but the durable demand sits in the underlying problem, anchor the recommendation in the underlying problem.
20. Channel strategy and hot-content tactics can inform entry angle, but they cannot replace topic validity.
21. A recommended course topic must be able to support at least three meaningful lessons without filler.
22. The material must contain a self-consistent teaching spine, not just isolated observations or scattered inspiration.
23. The report must judge whether the source contains enough distinctive viewpoints, methods, cases, experience, or exercises to make the course teachable.
24. Competitor comparison must test not only market positioning but also whether the material has a real teachable kernel that stands apart.
25. If the material can support a topic hook but not enough knowledge substance, downgrade the product form instead of forcing a course.

## Common Failure Modes

### Making the topic larger than the material

- Failure: a narrow capability gets packaged as broad market authority
- Fix: pull the topic back to the real audience, problem, and scenario the material actually covers

### Overstating the author

- Failure: scattered experience gets upgraded into full methodology
- Fix: call it a framework only when the material shows repeatable structure, cases, or stable judgment logic

### Using an audience that is too broad

- Failure: “anyone who wants to improve”
- Fix: specify at least two of the following: role, stage, current blockage

### Confusing need with willingness to pay

- Failure: assuming everyone with the pain is a viable customer
- Fix: test both problem intensity and product-form willingness to pay

### Mistaking discussion heat for business opportunity

- Failure: “people talk about it, therefore it should be a course”
- Fix: check saturation, alternatives, and whether the material has real replacement value

### Mistaking keyword growth for durable demand

- Failure: one rising term becomes a long-term course thesis
- Fix: compare 90-day and 12-month windows, and test the underlying problem terms

### Borrowing competitor language as if it were your own edge

- Failure: copying a successful framing that the material cannot support
- Fix: only reuse a framing if the material supports the promise, audience, and proof behind it

### Turning content validation into business validation

- Failure: public discussion is treated as proof of conversion
- Fix: content validation only proves that the topic deserves deeper testing

### Treating a course as the default product form

- Failure: anything teachable becomes a course by default
- Fix: test whether the better product is a lighter content product, tool, service, or downgrade

### Confusing a good topic sentence with a teachable course

- Failure: the title sounds strong, but the material cannot support three solid lessons
- Fix: audit knowledge points, methods, cases, and exercises before approving the course form

### Mistaking unique wording for real differentiation

- Failure: the report claims a differentiated angle, but competitors already cover the same substance
- Fix: compare the material's actual viewpoint, experience, method, and examples against competing offers

### Writing emotional buying logic in vague language

- Failure: “users are anxious and want growth”
- Fix: state the real purchase tension, such as fear of wasting time, fear of buying fluff, or fear of unusable output

## Suggested Scoring Dimensions

Use these as decision aids, not as a fake precision system:

- `source_strength`
- `audience_clarity`
- `market_timing`
- `solution_gap`
- `unique_value`
- `deliverability`
- `payment_fit`
- `validation_strength`
- `trend_stability`

Optional weighted scorecard for sharper decision-making:

- `source_strength`: 25
- `user_problem_fit`: 20
- `market_opportunity`: 20
- `differentiation_strength`: 20
- `commercialization_feasibility`: 15

Use this only when a weighted score genuinely helps the decision. Do not hide weak reasoning behind a neat total.

## Decision Status Definitions

For approval or resource-allocation outputs, add a formal decision status.

### `GO`

- a clear minimum sellable topic exists
- no hard gate is broken
- audience, problem, proof, delivery, and market stage align well enough to proceed
- the material can support a coherent course with enough substance for at least three meaningful lessons

### `HOLD`

- the direction may be good, but critical evidence is still missing
- common gaps include weak proof, unclear payment logic, or weak market evidence
- use this when the topic looks viable but lesson-level substance is still uncertain

### `REWORK`

- the material may support something real, but the current packaging is wrong
- likely fixes include changing the audience, lowering the promise, changing the product form, or narrowing the angle
- use this when the material has value but needs a smaller or more specific course scope to become teachable

### `NO-GO`

- no credible minimum sellable topic exists
- or the space is saturated and the material has no edge
- or buyers are clearly more likely to choose a tool or service
- or the material cannot support the delivery promise
- or the material cannot support enough coherent lessons to justify a course

## Risk Output Standard

Decision-oriented reports should explicitly score at least:

- `over_competition_risk`
- `insufficient_material_risk`
- `weak_differentiation_risk`

For each risk, include:

- severity
- trigger
- supporting evidence
- whether it changes the decision status

## Tone Rules

- write in judgment sentences, not motivational filler
- state boundaries clearly
- earn the commercial recommendation through coherence, not enthusiasm

# Input Contract

## P0 Required Inputs

At minimum, the analysis needs:

- core source material: transcripts, articles, outlines, notes, interviews, case writeups, or fragmented documents
- source ownership: whose experience, viewpoint, or method the material represents
- grouping or order across multiple documents, so different themes do not get merged by accident

Without P0, do not finalize a course-topic recommendation.

## P1 Strongly Recommended Inputs

- author background: role, experience, long-term expertise, representative outcomes
- proof assets: case studies, numbers, user feedback, before/after evidence, screenshots, work samples
- preferred market: geography, language, B2C vs B2B
- product constraints: format, duration, price band, whether coaching/community/training is in scope
- analysis mode: `A-material-only` or `B-market-scan` (`B-market-scan` is the default)
- research boundaries: which countries, platforms, and time windows should be prioritized

These inputs materially affect audience selection, promise ceiling, and differentiation strength.

## P2 Optional but High-Value Inputs

- existing target-customer hypotheses
- known competitors or reference courses
- explicitly forbidden directions
- already validated conversion angles or known failed angles
- seed keywords, user phrasing, or search phrases
- known effective channels
- existing validation signals: inquiry logs, sales calls, comments, high-performing content, landing-page data

## Language and Market Inputs

If the user's instruction language implies a market focus, the research scope should reflect that.

Rules:

- the final report should follow the user's instruction language unless the user says otherwise
- the market scan must include, and should prioritize, countries and markets aligned with the instruction language
- if the topic is cross-border, include the instruction-language market plus the most commercially relevant adjacent market

## Preprocessing Rules

- deduplicate the source set and preserve version relationships
- separate facts, cases, opinions, methods, and spoken texture before analysis
- record missing information explicitly rather than filling it in by guesswork
- assign `source_ref` IDs before analysis
- create a lightweight `trace_map`: `source_ref -> file / section / excerpt`

## Required Extraction Fields

Every run should extract at least:

- `theme_cue`: core themes and recurring claims
- `audience_cue`: audience, role, stage, baseline ability
- `problem_cue`: pain points, failure cases, blocked moments
- `outcome_cue`: outcomes the material can truly support
- `proof_cue`: cases, experience, numbers, methods, assets
- `boundary_cue`: limits, exclusions, assumptions, promise ceiling
- `language_cue`: recurring phrasing, tone, judgment style, voice markers
- `keyword_cue`: theme words, problem words, result words, alternative phrasing
- `platform_cue`: which platforms or formats the material naturally fits
- `monetization_cue`: whether the material leans toward course, community, tool, service, or pure content validation
- `evidence_ref`: the `source_ref[]` that supports each extracted cue

For stronger decision reports, target segments should also capture:

- `current_state`: what is true before the course
- `job_to_be_done`: the progress the user is trying to make
- `purchase_motivation`: why this person would actively buy
- `willingness_to_pay`: high / medium / low / unknown
- `budget_band`: rough spending range if visible
- `current_alternatives`: what the user is likely to buy or use instead

## Missing-Information Handling

If author-proof material is missing:

- you may say the topic is plausible but differentiation is weak
- you may not invent authority, years of experience, or proprietary methodology

If audience evidence is missing:

- you may propose 2-3 candidate audience groups
- each candidate group still needs a material-based reason

If market evidence is missing:

- downgrade the conclusion to “needs market validation”
- do not make strong opportunity claims
- if the task is a go/no-go decision, default toward `HOLD`, not `GO`

If keyword or competitor evidence is missing:

- generate an initial search set from the material itself
- explicitly note that SEO / trend / channel confidence is lower

If content-validation evidence is missing:

- you may judge topic plausibility
- you may not upgrade the conclusion to “validated to sell”

# Market Analysis

## Purpose

The point of market analysis is not to prove that “this will sell.” The point is to constrain topic selection with real public evidence:

- what the market is currently discussing
- how far user understanding has already developed
- what kinds of solutions dominate the space
- whether the source material still has a meaningful opening

Market analysis exists to discipline the recommendation, not to invent a marketing angle for weak material.

## Research Scope

You must research current public signals.

Default scope:

- prioritize the last 90 days
- extend to 12 months for mature or slower-moving topics

Signal categories to cover:

- discussion signals: articles, videos, posts, podcasts, communities, Q&A
- product signals: courses, bootcamps, consulting offers, tools, memberships
- demand signals: questions, complaints, “how do I,” “template please,” “example please,” “I bought this but still cannot do it”

Always record:

- concrete dates
- links
- the specific market observed

## Language-Aligned Market Rule

The market scan must reflect the user's instruction language.

Rules:

- if the user asks in Chinese, include and prioritize Chinese-language markets
- if the user asks in English, include and prioritize English-language markets
- if the topic is international, include the instruction-language market plus the most commercially relevant adjacent market
- do not run a language-mismatched market scan by default

## Default vs Optional Depth

The default research level is `core-scan`:

- lifecycle judgment
- competitor and solution mapping
- real user pain and unmet demand
- whether the source material fills a real gap

Add deeper layers only if necessary:

- `sizing-scan`
- `seo-scan`
- `trend-scan`
- `channel-scan`
- `validation-scan`

Do not expand every topic-selection task into a full commercial research project.

## Demand Density vs. Market Size

At the topic-selection stage, demand density matters more than fake precision on market size.

Check:

- does the same problem appear across platforms
- is there stable paid supply, not just free discussion
- will users pay for a course rather than a tool or service
- do pricing bands and update frequency suggest a stable category
- does the same underlying problem persist across multiple audience groups

Allowed labels:

- `small`
- `medium`
- `large`
- `unknown`

Always clarify that this is a public-signal estimate, not a precise market model.

## Keyword and Trend Analysis

Do not scan a topic with a single keyword.

At minimum, search across:

- core topic terms
- problem terms
- result terms
- alternative phrasing and competitor phrasing

Trend checks should compare:

- 30 days
- 90 days
- 12 months

Look for:

- `event-spike`
- `seasonal`
- `steady-demand`
- `declining`
- `reframed-demand`

If the visible heat is only in the framing while the underlying problem is stable, say so explicitly.

## Lifecycle Judgment

Use one primary stage, with an optional secondary stage if needed.

### `information-explosion`

- lots of broad discussion
- shallow understanding
- high visibility, weak differentiation

### `segmented-understanding`

- the broad idea is accepted
- different user groups now have different questions
- this is often the best stage for a sharp segment play

### `methodology-phase`

- the market wants repeatable process, not just explanation
- framework-heavy solutions appear
- weakly structured material tends to fail here

### `toolification-phase`

- tools, templates, or semi-automation increasingly replace manual work
- users want speed and convenience more than more theory
- pure content products face replacement risk

### `red-ocean`

- repetitive headlines
- repetitive promises
- pricing pressure
- low trust in generic offers
- buyers only move for stronger proof, narrower fit, or much better value

## Competition Types

Judge not only stage, but also what type of competitor is strongest.

### Content competitors

- courses
- tutorials
- cohorts
- training camps
- knowledge products

### Tool competitors

- SaaS
- AI writing tools
- templates
- workflow products

### Service competitors

- consultants
- ghostwriters
- done-for-you services
- coaching or implementation help

The report should say:

- which type dominates
- why a course still has room, or why it does not

## Competitor Analysis Standard

For each serious competitor or substitute, answer:

- what problem it solves
- who it serves
- what it promises
- how it delivers
- why buyers choose it
- what it does not solve
- what room remains for this topic

Do not limit competitors to “courses like mine.” Include tools, free content, training camps, platform features, and service substitutes.

## When a Market Is “Too Crowded”

You can classify a space as crowded or close to red ocean when several of these are true:

- top and mid-tier competitors make nearly the same promise
- keywords and framing are repetitive
- users describe offerings as “all the same”
- price pressure is visible
- tools or services are replacing course demand
- the material has no extra proof, sharper audience, or structural advantage

Useful labels:

- `under-contested`
- `segmentable`
- `red-ocean`
- `tool-replaced`

## Solution Mapping

For each major solution type, note:

- `solution_type`
- `target_user`
- `promise`
- `delivery`
- `strength`
- `weakness`
- `evidence_level`
- `replacement_risk`

This is where you explain what the market is already solving, and what still remains unsolved.

## Opportunity Logic

A real opportunity usually looks like one of these:

- many people talk about the theme, but the audience segmentation is still weak
- the market explains the topic, but the material offers a more executable path
- the market overpromises results, while the material offers stronger process credibility
- the market is fragmented, but the material contains a coherent system
- the market is tool-heavy, but users still need judgment and refinement

These do **not** count as real opportunities:

- merely changing the headline
- making the topic broader
- pretending weak material can compete with strong proof-based offers
- targeting a group that needs help but is more likely to buy a tool or service

## Real User Pain

Prioritize user language, not creator language.

Look for:

- repeated questions
- complaints and failure reports
- requests for templates, cases, examples, or process
- “I bought this and still can’t do it”
- “the tool is fast, but the output is unusable”

Map those pains to:

- `trust`
- `relevance`
- `executability`
- `value`

## Channel and Content Pattern Analysis

Only do this when the user asks for channel strategy.

Check:

- active platforms and positioning
- title structures and keyword clusters
- what content pulls attention vs. what content converts
- whether the market wins through cases, opinions, tutorials, tests, or comparisons

The purpose is to help select the market entry angle, not to copy anyone's content.

## Content Validation

Content validation does not mean “guaranteed to sell.” It means there is enough public signal to justify going deeper.

Useful labels:

- `validated`
- `partially-validated`
- `weakly-validated`
- `unvalidated`

At minimum, validate against:

- recurring problems inside the source material
- recurring market demand
- stable supply of comparable offers
- visible dissatisfaction that the material could credibly address

## Emotional Purchase Logic

Beyond rational need, identify the emotional purchase driver:

- `trust`
- `relevance`
- `executability`
- `value`

The report should say which matters most for each target user, and whether the source material can support it.

# Output Contract

## Required Outcome

The result must end in one of two states:

- viable: one recommended topic plus two to three backup directions
- not viable: a formal `not-viable` conclusion, optionally with downgrade paths

## Required Fields for Every Topic Direction

- `analysis_mode`: `A-material-only` or `B-market-scan`
- `viability_status`: `recommended`, `alternative`, `downgraded`, or `not-viable`
- `topic_type`: cognitive, method, case, execution, outcome-led, or hybrid
- `title`: the course title
- `one_line_pitch`: who it is for and what problem it solves
- `source_boundary`: what the topic is strictly based on, and what it does not include
- `source_evidence`: the supporting cases, methods, excerpts, and proof, with `source_ref[]`
- `target_users`: one to three audience segments, each with stage, pain point, and buying reason
- for stronger reports, each target segment should also explain current state, job to be done, purchase motivation, willingness to pay, and current alternatives
- `non_buyers`: who should not be targeted, and why they are unlikely to buy
- `lifecycle_stage`: the current market stage
- `competition_type`: content, tool, service, or a combination
- `market_summary`: current heat, dominant solution types, and characteristic competitive patterns
- `opportunity_gap`: why the topic still has room, or why room is limited
- `unique_value`: the real differentiators supported by the material
- `buying_driver`: the dominant buying logic, explained through `trust / relevance / executability / value`
- `promise_ceiling`: the strongest promise level the material can honestly support
- `risk_note`: why the topic may not sell, or why it may sound empty in delivery
- `course_readiness`: whether the material is strong enough to support a course rather than only a topic angle
- `lesson_spine`: the minimum coherent lesson structure the material can support; use `insufficient` if it cannot yet support a real course
- `missing_info`: what additional evidence would make the conclusion stronger

## Required Additions for Decision / Approval Reports

- `decision_status`: `GO / HOLD / REWORK / NO-GO`
- `decision_reason`: why this is the right decision rather than a more optimistic or more conservative one
- `top_risks`: at least three risks, typically including competition, evidence weakness, and differentiation weakness
- `competition_matrix`: a matrix that explains what major competitors or substitutes solve, who they serve, why people buy them, and what gap remains
- `claim_evidence_map`: a claim-level map to `source_ref[]` and `market_ref[]`
- `scorecard`: an optional weighted scorecard if the user wants a sharper go / hold / rework judgment
- `price_hint`: recommended pricing or price band; use `unknown` if evidence is insufficient
- `mvp_scope`: the smallest launchable version of the course
- `validation_plan_7d`: a seven-day validation plan for testing demand, messaging, and sample content
- `teachable_kernel`: the distinctive viewpoints, methods, cases, experience, or insight density that make the topic teachable versus generic competitors

## Optional Additions for Deeper Market Work

Add only when the user asks for more depth:

- `demand_density`: `small / medium / large / unknown`
- `market_size_note`: a rough market-size note grounded in public signals
- `trend_pattern`: `event-spike / seasonal / steady-demand / declining / reframed-demand`
- `saturation_level`: `under-contested / segmentable / red-ocean / tool-replaced`
- `seo_gap`: public-content keyword and positioning gap
- `channel_hint`: likely entry channels and content angles
- `content_validation_status`: `validated / partially-validated / weakly-validated / unvalidated`
- `validation_basis`: the market evidence behind the validation judgment

## Required Additions for the Recommended Direction

The recommended topic must also explain:

- `why_this_one`: why it wins over larger, hotter, or more obvious directions
- `fit_reason`: how it matches the material, the user problem, and the market stage at the same time
- `go_to_market_hint`: the best packaging angle for entering the market
- `growth_shape`: whether it behaves more like a trend-led topic, a durable demand topic, or a hybrid
- `expected_curve`: whether the direction is likely to spike, compound steadily, or follow a hybrid curve
- `upside_case`: what has to go right for the topic to work especially well
- `downside_case`: what could make the topic underperform
- `course_readiness_reason`: why the material is sufficient to support a course, including the minimum three-lesson threshold when applicable

## Human-Facing Report Rules

If the report is meant for humans rather than downstream systems:

- the main and backup directions must differ meaningfully in audience, angle, teaching objective, or product form
- competitor analysis cannot stop at links; it must explain what problem each product actually solves
- the report should explicitly state whether the material is topic-ready only or genuinely course-ready
- prefer natural language over field-name-heavy presentation
- if both traceability and readability matter, produce:
  - `selection_report.md`: structured version
  - `decision_brief.md`: readable version

## `not-viable` Output

If no competitive topic can be justified:

- use `viability_status: not-viable`
- state whether the failure is due to content, audience, market, or delivery weakness
- include downgrade options where appropriate

## Title Rules

A title should express at least two of the following:

- audience
- problem
- result

It must not:

- exceed the material boundary
- promise results the evidence cannot support
- become abstract or “high-level” at the cost of specificity

## Recommended Output Order

1. conclusion first
2. backup options and why they rank below the winner
3. `not-viable` or downgrade conclusion if needed
4. market judgment: stage, competition, dominant solutions, room to win
5. deeper scans only if required
6. real material-based advantages
7. risks and missing evidence

## Reusable Output Files

When useful, also generate:

- `topic_candidates.json`
- `market_scan.md`
- `source_evidence_map.json`
- `selection_report.md`
- `decision_brief.md`